OpenAI and Anthropic Slash AI Model Prices 25% as Chinese Rivals Moonshot and DeepSeek Gain Ground

OpenAI and Anthropic Slash Mid-Tier AI Model Prices

The AI price war has escalated dramatically as OpenAI and Anthropic implement aggressive price cuts to defend their market share against surging Chinese AI rivals1

1

. OpenAI slashed prices for GPT-5.6 Luna, its fastest and most affordable model, by 80%—dropping from $1 to $0.20 per million input tokens and from $6 to $1.20 per million output tokens2

2

. Anthropic launched Claude Opus 5 at $5 per million input tokens and $25 per million output tokens, positioning it at half the price of its flagship Fable 5 model1

1

. These moves helped decrease prices that customers are paying for models from leading US labs by almost 25% since mid-July, according to Silicon Data's token price index2

2

.

The price reductions mark a strategic shift for US AI groups that make proprietary closed models and have historically competed heavily on performance rather than cost1

1

. Anthropic went further by reversing a scheduled price increase for its Claude Sonnet 5 model, canceling a planned rise from $2 to $3 per million input tokens that was set to take effect in September4

4

. OpenAI also cut GPT-5.6 Terra, its mid-tier model, by 20% from $2.50 to $2 per million input tokens while leaving its flagship GPT-5.6 Sol unchanged at $54

4

.

Chinese AI Rivals Moonshot and DeepSeek Capture Enterprise Customers

Rising AI usage costs have pushed companies to impose caps on AI usage or test cheaper alternatives, with major enterprises like DoorDash and Airbnb publicly stating they have started using Chinese-made models to rein in bills2

2

. Chinese developers including Moonshot and DeepSeek are making significant inroads with users from Silicon Valley to Europe by offering increasingly capable open models that can be freely downloaded and tweaked by developers1

1

. This shift has coincided with a flurry of releases from Chinese labs that have narrowed the performance gap with leading US models, raising concerns in the US tech industry that American developers could lose customers even as they spend heavily to maintain their technological edge2

2

.

The competitive pressure intensified as Anthropic and OpenAI shift some enterprise customers away from flat subscriptions toward usage-based billing, under which companies pay according to the computational resources they consume1

1

. Venture capital firm Andreessen Horowitz has tracked this phenomenon since 2024, coining the term "LLMflation" to describe a roughly 10-fold price decline every year for a model of fixed capability4

4

.

Cost Per Task Reveals Complex Pricing Reality

Headline token prices don't provide a straightforward comparison between AI models, as more capable models can sometimes complete a task using fewer tokens or with fewer attempts2

2

. Research from AlphaSense found that OpenAI's GPT-5.6 Sol delivered answers with roughly 20% higher quality while costing about 13% less than Moonshot's Kimi K3 on a median basis when analyzing 246 financial analysis tasks3

3

. Anthropic's Opus 4.8 performed even better, generating responses that scored around 13% higher in quality at roughly half the overall cost of Kimi K33

3

.

Artificial Analysis, which benchmarks models across areas including math, science, coding, and reasoning, found Anthropic's Opus 5 at medium effort delivered similar performance and cost per task to Moonshot's Kimi K3 at max effort1

1

. OpenAI's GPT-5.6 Luna at max effort performed similarly to DeepSeek's V4 Flash at max, but cost just under twice as much per task1

1

. Most models can operate at different effort settings, which vary the computing power used to answer a question and can affect both performance and the ultimate cost of completing a task2

2

.

US Labs Defend Premium Tier While Cutting Mid-Range Prices

Mantas Lukauskas, AI tech lead at Hostinger, a website hosting provider that has used large language models since 2020, noted that prices for the very best models were "flat to rising"1

1

. He characterized the recent pricing changes as the "first real test" of whether groups such as Anthropic and OpenAI can protect the cost of their most advanced offerings, stating: "The US labs have cut the middle and are defending the top"2

2

. At OpenAI, the spread between its cheapest and most expensive models widened to 25-to-1 from 5-to-1 in a single announcement4

4

.

The price cuts arrive as OpenAI and Anthropic plot initial public offerings at trillion-dollar valuations while investors seek evidence that the industry's vast spending on AI can generate returns1

1

. AlphaSense argues businesses should evaluate the total cost of completing a task alongside the quality of the output, suggesting the most cost-effective strategy may be using a multi-model strategy that employs a routing system to automatically assign different parts of a query to different models3

3

. As enterprise AI adoption accelerates, the pattern mirrors broadband internet, where falling prices at the bottom of the market expanded access more than any improvement to premium tiers4

4

.